AI-powered interactive HTML experiences that put students inside every NGSS practice — not just reading about science, but actually doing it.
The NGSS Science & Engineering Practices define what scientists and engineers actually do. But traditional instruction often reduces them to vocabulary on a rubric. SEPsim exists to change that — using AI to generate rich, interactive simulations that drop students directly into each practice.
Every module on this site was built with Claude or another large language model — and you can build your own in minutes. No coding degree required.
Click any practice to see how interactive simulations can bring it to life in your classroom — and what to ask AI to build for you.
Scientists ask questions about the natural world; engineers define problems that must be solved. This practice is the ignition switch for inquiry — but students need scaffolding to ask testable questions rather than Googleable ones.
Models are the language of science — diagrams, simulations, mathematical representations, physical mockups. Students must learn to build models AND critique their limitations.
Students must design investigations — not just follow lab protocols. They need to identify variables, choose appropriate tools, and decide what data to collect before a single measurement is taken.
Raw data is meaningless until analyzed. Students must identify patterns, compute statistics, visualize trends, and separate signal from noise — skills that transfer far beyond any science class.
Science without math is just vibes. This practice asks students to use quantitative reasoning — not just plug-and-chug, but understand what the numbers mean and how models behave at scale.
The pinnacle of science: building a coherent explanation grounded in evidence. And the pinnacle of engineering: proposing a solution that actually works. Both require the CER framework and strong causal reasoning.
Science is not a collection of facts — it is an ongoing argument settled by evidence, not authority. Students must learn to make, critique, and revise scientific arguments using data.
Scientists read primary literature. They evaluate source credibility. They communicate findings to different audiences. In an age of misinformation, this practice may be the most urgent of all.
You don't need to know how to code. You need to know your students, your content, and how to ask a good question — which, conveniently, is SEP 1.
Identify which SEP your students need to practice. What does that practice look like in action in your content area? What phenomenon or context fits your current unit?
Tell Claude what you want: the SEP target, your topic, your students' level, and what kind of interaction you envision. Be specific about inputs, outputs, and feedback type.
Review the generated HTML in your browser. Ask Claude to adjust the difficulty, add scaffolding, change the topic, or restyle the interface. This is design thinking, not one-and-done.
Upload to your class server or LMS. Add AI grading via the Anthropic API, connect to Google Sheets for logging, or embed in Schoology via SCORM. Students get feedback instantly.
Pick one or more practices, choose your interaction types, describe your course and topic — then let the AI write you a detailed, ready-to-use simulation prompt in seconds.
"Students learn science by practicing science — not by reading about it. Every SEP simulation is a minute less lecture and a minute more doing."— The core philosophy of SEPsim
Traditional interactive resources take weeks to build and rarely match your specific curriculum. AI changes the equation.
A full interactive module that would take a developer days can be built and refined in under an hour. When a simulation isn't working pedagogically, you describe what's wrong and regenerate — no code archaeology required.
Generic simulations use generic data. AI lets you set your simulation in your region — your watershed, your local air quality data, your community's energy mix. That specificity drives engagement and real transfer.
Wire in the Anthropic or OpenAI API and your simulation can give students instant, individualized feedback on their reasoning — not just right/wrong, but why. That's a tutor in every seat.
Need a version with more scaffolding for struggling learners? An extended challenge for early finishers? Generate both from the same base prompt with a single sentence of instruction. No extra prep time.
When you specify a SEP in your prompt, Claude aligns the activity to that practice by design — not as an afterthought. Include the performance expectation code and it threads the needle between the SEP, the DCI, and the crosscutting concept.
Every simulation lives in a single HTML file. Students open it in a browser. It works on school Chromebooks, iPads, and seven-year-old Windows laptops. No accounts, no logins, no IT tickets.